Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: two-way error component model

From   Mark Schaffer <>
To, Svetlana Mira <>
Subject   Re: st: two-way error component model
Date   Thu, 24 Feb 2005 19:53:15 +0000 (GMT)


Quoting Svetlana Mira  <>:

> Dear All,
> I have a problem understanding how Stata computes the results for a
> two-way error component model (xi: reg y i.idfirm i.idtime x1) and
> hopefully somebody will be able to help me with this. In the case of
> a fixed effects model, we can obtain the results in Stata either by
> using xtreg, areg or simply based on the demeaned values using 
> regress option. Can we apply the last option (demeaning) for the two-
> way error component model? (i.e., can we run a two-way error 
> component model based on the mean differences)? If so, what means 
> shall we extract?
> For instance, if we run xi: reg y i.idfirm i.idtime x1 the obtained
> results are not equal to the results based on 
> regress y x2, where x2=x1-x(meanidfirm)-x(meanidtime)      
> Or as suggested by Green (1997)
> regress y x3, where x3=x1-x(meanidfirm)-x(meanidtime) + x(firm-year
> specific mean).

I am pretty sure that if you check Greene, you'll find that this is the
formula for obtaining the 2-way error components estimate for a balanced
panel using mean deviations.  If the panel is unbalanced, the mean-deviation
approach requires calculations that are somewhat more involved.  Baltagi's
book "Econometric Analysis of Panel Data" has a discussion of this.

Hope this helps.


> Does Stata use a specific procedure to run a two-way error component
> model different from the ones mentioned above? Any suggestions or 
> comments are more than appreciated.
> Many thanks for your help, 
> Svetlana
> *
> *   For searches and help try:
> *
> *
> *

Prof. Mark Schaffer
Director, CERT
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3294


This message is subject to 


*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index